A Theoretical Question: EQ vs mic design

  • Thread starter Thread starter mcolling
  • Start date Start date
M

mcolling

New member
OK, I know that mic choice likely the most important aspect of the signal chain, and that the sound of a poor mic cannot be "fixed in the mix". This is the conventional wisdom, and I agree based on experience.

However, I want to question this assumption for one moment. Theoretically, couldn't a truly accurate mic be used for recording, with all the pleasant colouration and eq of, say, a Neuman U87 added during mixdown?

I know that this concept has been attempted before with the Anteres Mic Modeller, and (according to everyone) it hasn't been successful. But is there a theoretical reason why, or was it just a flawed execution?
 
A mic is more than its on-axis frequency response. But even if you had a perfectly flat, fast transient, ultralow distortion binaural setup that allowed you to create any polar pattern and frequency response via mic modeling . . . um, a DJ still wouldn't be a musician :D

I think the main problem was people trying to take a mic with poor high frequency response and turn it into a U87. If the source signal isn't there in the recording, you don't know what to recreate.
 
No.

For one thing, a large part of the way a mic sounds is a part of the capsule. The capsule does not introduce phase discrepancies. An EQ does, and boy howdy does it.

Secondly, a mic picks up room ambience differently than it does the direct signal, and differently at every frequency. At 20 Hz, EVERY mic is omni directional. At 20,000 Hz, they are all extremely directional. But they all handle the area in the middle differently. THAT, as much as anything, is what makes up the sound of a mic.

And then there is the rarely mentioned (or at least, it used to be rarely mentioned) aspects of response/rise time. A large diaphragm mic reacts to sound much slower than a small diaphragm mic. And of course, they both respond differently at different frequencies. Try and get an EQ to recreate THAT.


Light

"Cowards can never be moral."
M.K. Gandhi
 
mshilarious said:
um, a DJ still wouldn't be a musician :D


MOST DJ's may not be musicians. There are exceptions.

And no, I don't have any names, as I don't really pay much attention to what they are doing, but I have heard it.


Light

"Cowards can never be moral."
M.K. Gandhi
 
Isn't that kinda' like saying: "A boat isn't a house. But I know of some people who live in their boat." ?
 
sometimes i can't find the little man in the boat.....



wade
 
Light said:
No.

For one thing, a large part of the way a mic sounds is a part of the capsule. The capsule does not introduce phase discrepancies. An EQ does, and boy howdy does it.

Secondly, a mic picks up room ambience differently than it does the direct signal, and differently at every frequency. At 20 Hz, EVERY mic is omni directional. At 20,000 Hz, they are all extremely directional. But they all handle the area in the middle differently. THAT, as much as anything, is what makes up the sound of a mic.

And then there is the rarely mentioned (or at least, it used to be rarely mentioned) aspects of response/rise time. A large diaphragm mic reacts to sound much slower than a small diaphragm mic. And of course, they both respond differently at different frequencies. Try and get an EQ to recreate THAT
Good answer!! Perfect. That helps too, mshilarious.

So, the next question is, how important are the transient response and the response/rise time to the overall sound of a mic?

And, ya, the old DJ Debate. Some are undeniably musicians, haha :D. For instance Mix Master Mike- that man is a virtuoso on par with anyone I have ever seen, be them a pianist, violinist or guitarist. But the dude's spinning trance albums... not so much.
 
Last edited:
mcolling said:
Good answer!! Perfect.

So, the next question is, how important are the transient response and the response/rise time to the overall sound of a mic?

It's very critical.

On Light's point, no an EQ cannot recreate a faster-than-recorded transient response. Even modeling algorithms have their limits. But if you had a theoretical mic that "perfectly" recorded the entire soundstage (at least a two dimensional soundstage, presuming we are headed for a stereo mix), then it should be fairly trivial for a modeling algorithm to 'muddy up' the pristine signal, or change the polar response, slow down transient response, whatever.

That would require an expensive pair of mics though. In other words, it's pretty easy to make high-quality mics sound lo-fi, but the customers of the mic modelers want to make a crappy mic sound like a good one. That just ain't gonna happen.

And, ya, the old DJ Debate. Some are undeniably musicians, haha :D. For instance Mix Master Mike- that man is a virtuoso on par with anyone I have ever seen, be them a pianist, violinist or guitarist. But the dude's spinning trance albums... not so much.

The "Is a DJ a musician" question is rhetorical, not in that no DJs are musicians, but the question itself is pointless. It's a running joke on the site.
 
no, mics deal with sound pressure and the way a mic reacts to that pressure would not be possible to re-create with eq.

I think the same could be said for speaker drivers. A good speaker sounds diffrent that a bad speaker and an eq would not make the bad speaker sound like the good speaker.
 
mshilarious said:
...


then it should be fairly trivial for a modeling algorithm to 'muddy up' the pristine signal, or change the polar response, slow down transient response, whatever.


...


umm... How, exactly?

I mean, we are talking not just about frequency response. Remember, the frequency response differs drastically depending on where the sound is coming from in three dimensions. A stereo signal is only a ONE dimensional signal - left/right. How, exactly do you expect any modeling device to figure out the rest of the spatial information? It just is not there.


Light

"Cowards can never be moral."
M.K. Gandhi
 
Light said:
umm... How, exactly?

I mean, we are talking not just about frequency response. Remember, the frequency response differs drastically depending on where the sound is coming from in three dimensions. A stereo signal is only a ONE dimensional signal - left/right. How, exactly do you expect any modeling device to figure out the rest of the spatial information? It just is not there.

OK a surround signal then. My point is that sound pressure at a given location can be measured by amplitude, and when you add multiple points, direction. OK, it takes three points to make a plane.

Beyond that it's just a question of the quality of the transducers, but that's a practical problem, not a theoretical problem.

Responding to boomtap: DSP gets a lot more sophisticated than EQ, and I never claimed that DSP could make a bad transducer sound good. In fact I specifically rejected that.

I guess in my attempt at DJ humor, it was missed that I agreed with the premise.
 
Theoretically, couldn't a truly accurate mic be used for recording, with all the pleasant colouration and eq of, say, a Neuman U87 added during mixdown?

I know that this concept has been attempted before with the Anteres Mic Modeller, and (according to everyone) it hasn't been successful. But is there a theoretical reason why, or was it just a flawed execution?

I don't think audio science is anywhere near there yet. Sure, you can measure and describe a mic's response in various ways, but I think that recording is more alchemy - or art - than science.

Tim
 
Light said:
For one thing, a large part of the way a mic sounds is a part of the capsule. The capsule does not introduce phase discrepancies. An EQ does, and boy howdy does it.

A phase-linear digital EQ doesn't. I think this is pretty much beside the point. It won't sound a jot closer to the attempted mic with a linear EQ versus a regular one.

And then there is the rarely mentioned (or at least, it used to be rarely mentioned) aspects of response/rise time. A large diaphragm mic reacts to sound much slower than a small diaphragm mic. And of course, they both respond differently at different frequencies. Try and get an EQ to recreate THAT.

Now, Anatres Mic Modeller may not be a shining star, but it DOES model transient response, and apparently quite accuractely. It can even quicken up a slower response accurately. I believe it does this through some sort of convolution. I don't think the answer lies here, either.
Please note I am going off the only available paper I could find written on Mic Modeller's transient response, which is by the Antares engineers themselves, so it is obviously biased.
 
Timothy Lawler said:
I don't think audio science is anywhere near there yet. Sure, you can measure and describe a mic's response in various ways, but I think that recording is more alchemy - or art - than science.
Tim

Hogwash. Every physical system can be accurately described in terms of science and logic. If it's not currently working, then one of two things are going on:

1) You aren't looking at all the relevant parameters, or the parameters that you have aren't accurate enough, or

2) The data you're feeding the parameters is insufficient/inaccurate.


As far as #2 goes, we all know there is significant variability among various samples of one model of microphone. This could contribute, among many other things.
 
Every physical system can be accurately described in terms of science and logic.

bleyrad, would you call the brain a physical system?

Tim
 
Timothy Lawler said:
bleyrad, would you call the brain a physical system?
Tim
This is gonna get way off topic, but I'll bite.

In one sense, yes for sure.
In another sense, yes and I don't know.

Let me explain. In the first sense I mean that even if there is something non-computationally-based going on in terms of our consiousness, it holds the potential to be explained. In other words, even if we have a soul (and I must stress that no one really knows either way, and current evidence is pointing towards our consiousness simply being a grand illusion by the sum of the brain's processes upon itself) this is potentially explainable. That is, if we had enough information about what was going on, we should be able to figure out what consiousness is and how this "soul" works. We don't currently have enough information, and we might NEVER have enough information, but that doesn't mean it's not a scientifically valid process. Everything in this universe IS, simply because it exists (or doesn't exist). Such is the nature of science.

In the second sense of my answer, I can back up my confident "yes" part by saying that we already know a great deal about how the brain works. It's essentially a very complex digital bioelectrical processor. Each neuron is a "transistor" which affects the potential of other connected neurons to either fire or not fire. The ends of the train lay at our sensory input, like sight and touch, and our motor outputs, such as movement and speech. Input-output. And somewhere in between is a huge amount of processes that, when put together, form the (possible) illusion known as consiousness. Note that as soon as you change any of the way these sensory processes and neurons fire or inhibit from firing, your perception of what consiousness is changes. I'm talking about mind-altering drugs, some of which our own body produces! All it takes are a few chemicals setting off neurons and changing the whole of the process to look at yourself and the world in a completely different way - in a sense, changing the nature of consiousness itself.
This is where the "I don't know" part of my answer comes from. I don't know, because no one knows, exactly what consiousness is or what causes it. I have my guesses, as does everyone else. What bothers me though is people who claim to know the answer for certain (which is a lot of people). The only wise thing to do right now is tolerate our uncertainty towards this matter. Anything else is self-deception. There is no evidence or proper information towards any other answer.
I have problems with faith-based answers because faith is such a finnicky thing and IMO not to be trusted. Everyone strongly, strongly feels that theirs is the correct answer, right to their very bones... yet many of these "answers" conflict deeply. I hate this example but it works: Hitler felt strongly that whites were the most superior human beings, strongly enough to massacre millions of those who weren't. We're talking about an even stronger conviction and certainty in his beliefs than many religious folk have in theirs today.
As soon as we apply very basic logic to his actions though, they come out wrong and irrational. Such is the case with all faith, unfortunately. The only thing it's done is hold us back, and it's only within the past few hundred years that we've moved beyond that into something far less biased (the scientific method). It's a pity that so many people are automatically distrusting of science, repeating the mistakes of generations and generations of anscentors with internally-based beliefs that sometimes stand in the way of true knowledge.
 
bleyrad said:
Hogwash. Every physical system can be accurately described in terms of science and logic. If it's not currently working, then one of two things are going on:

1) You aren't looking at all the relevant parameters, or the parameters that you have aren't accurate enough, or

2) The data you're feeding the parameters is insufficient/inaccurate.


As far as #2 goes, we all know there is significant variability among various samples of one model of microphone. This could contribute, among many other things.

Don't forget one simple thing: COMPLEXITY. Not every system CAN in fact be foreseen... There are systems so sensitive to parameter changes, that you'll have completely different result after a certain time, that the numerical rounding errors lead to completely different results when you switch from 32 Bit to 64 Bit floats. Those things are in fact VERY unlikely to be predicted correctly though you might think they are. (Example: the balancing of two sticks one above another seems to be a problem like that).

Nevertheless, I assume that you could simulate mics by using networks of simple measurement mics and enough knowledge of the other mics and the room etc...

aXel
 
I think someday we will use DSP to send signals direct to the human brain. But first we have to try it out on monkeys.
 
I think science relating to the physics of audio recording is very complex because it's coupled with the psychological aspects of our audio perception. And I'm not talking about metaphysics of the "soul" or "consciousness". Just the way we perceive musical sound.

A couple examples come to mind of complex systems that are understood in only a partial way. Psychiatry recognizes the fact that the brain is so complex that we don't know, except in a rather crude way, why certain medications work for some mentally ill patients and not for others with the same diagnosis. And weather prediction is correct how much of the time? I think many physical systems are just beginning to be understood. Science relating to complex systems has changed a lot recently in its belief that we really understand our world, relating especially to complex adaptive systems and certainly to systems affected by chaotic aspects.

Anyway, my point isn't that scientific study of audio is not worthwhile, just that it's a ways off from producing the results the OP was asking about.

Tim
 
I agree!

But it's a challenge that anyone with the right mindset could potentially overcome given enough time and dedication to understanding, is it not?
Perhaps one of us will be the one to create a mic modeller that really works.


I think if you develop that kind of advanced technolgy though, there are probably more important uses for it, even if you restrict yourself purely to the recording aspect.
 
Back
Top